=Paper=
{{Paper
|id=Vol-2807/paperM
|storemode=property
|title=The ARK Platform: Enabling Risk Management Through Semantic Web Technologies
|pdfUrl=https://ceur-ws.org/Vol-2807/paperM.pdf
|volume=Vol-2807
|authors=Ademar Crotti Jr.,Maryam Basereh,Yalemisew Abgaz,Junli Liang,Natalia Duda,Nick McDonald,Rob Brennan
|dblpUrl=https://dblp.org/rec/conf/icbo/CrottiBALDMB20
}}
==The ARK Platform: Enabling Risk Management Through Semantic Web Technologies==
https://ceur-ws.org/Vol-2807/paperM.pdf
Crotti Junior, A. et al. /
The ARK Platform: Enabling Risk
Management through Semantic Web
technologies
Ademar CROTTI JUNIOR b , Maryam BASEREH a , Yalemisew ABGAZ a ,
Junli LIANG c , Natalia DUDA c , Nick MCDONALD c , and Rob BRENNAN a,1
a ADAPT Centre, School of Computing, Dublin City University, Dublin, Ireland
b ADAPT Centre, School of Computer Science and Statistics, Trinity College Dublin,
Dublin, Ireland
c School of Psychology, Trinity College Dublin, Dublin, Ireland
Abstract. This paper describes the Access Risk Knowledge (ARK) platform and
ontologies for socio-technical risk analysis using the Cube methodology. Linked
Data is used in ARK to integrate qualitative clinical risk management data with
quantitative operational data and analytics. This required the development of a
novel clinical safety management taxonomy to annotate qualitative risk data and
make it more amenable to automated analysis. The platform is complemented
by other two ontologies that support structured data capture for the Cube socio-
technical analysis methodology developed by organisational psychologists at Trin-
ity College Dublin. The ARK platform development and trials have shown the ben-
efits of a Semantic Web approach to flexibly support data integration, making quali-
tative data machine readable and building dynamic, high-usability web applications
applied to clinical risk management. The main results so far are a self-annotated,
standards-based taxonomy for risk and safety management expressed in the W3C’s
standard Simple Knowledge Organisation System (SKOS) and a Cube data capture,
curation and analysis platform for clinical risk management domain experts. The
paper describes the ontologies and their development process, our initial clinical
safety management use case and lessons learned from the application of ARK to
real-world use cases. This work has shown the potential for using Linked Data to
integrate operational and safety data into a unified information space supporting
more continuous, adaptive and predictive clinical risk management.
Keywords. ARK Platform, Organisational Change, Risk Management.
1. Introduction
Managing clinical risk is an integral part of any organisation providing clinical services.
The Cube is an established methodology for analysing socio-technical systems which
offers a framework for managing such risks [1, 2, 3]. It is used to identify, assess, and
classify risks, and to plan, execute, and evaluate risk mitigation actions (projects). The
1 Corresponding Author: Rob Brennan; E-mail: rob.brennan@adaptcentre.ie.
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Crotti Junior, A. et al. /
Cube methodology helps domain experts building case studies which lead to an evidence
base that supports better practice and governance.
Organisations already collect and study (human-oriented) risk information in the
form of risk registers, annual safety reports and incident reports. They are also increas-
ingly overloaded with digital operational data that could be used to inform risk manage-
ment as well as operational management. Typically these two management cycles op-
erate independently, upon siloed data, and focus on qualitative analysis (risk) or quan-
titative analysis (operations in digitally transformed organisations like modern clinical
practice) respectively. These differences mean that it is hard to find machine-readable
rather than human-focused risk data and even harder to fuse it with structured opera-
tional data or predictive analytics from data science projects, despite the utility that these
sources may have for implementing a more dynamic and effective risk management sys-
tem. These disconnected activities also make it difficult to efficiently analyze and build a
machine readable evidence base that is suitable for modern automated or semi-automated
analytical approaches. Even though the Cube methodology supports such activities, its
deployment has been hampered by the lack of an integrated platform facilitating its use.
This paper describes the Access Risk Knowledge (ARK) platform for Cube-based
risk analysis and evidence base collection. ARK uses Semantic Web technologies to
model, integrate, and classify risk and socio-technical system analysis information from
both qualitative and quantitative data sources into a unified risk graph. This graph is
based on three new ontologies: a risk terminology expressed as a SKOS model, a Cube
analysis vocabulary, and an ARK project vocabulary. The ARK platform and ontologies
have been built to study this research question: ”to what extent can a Linked Data ap-
proach support semantic interlinking and interoperability between quantitative and qual-
itative risk data, extraction and representation of risk domain knowledge, and enhance
dynamic and integrated analysis of risk by supporting the Cube methodology?”.
ARK is built as a native web application in Node.js with a React user interface driven
from an Apache Jena triplestore based on our developed ontologies. The ontologies are
based on two years of iterative development between mixed teams of organisational psy-
chologists and knowledge engineers. By using these technologies, the ARK platform
provides a way to standardize social-technical analysis outputs into a self-describing
semantic graph that can include Cube structured questionnaire-based analysis, multi-
dimensional synthesis, or risk mitigation project-oriented views of the organisation under
study. It also allows for semantic interlinking and interoperability of diverse data sources
such as operational data or analytics. It facilitates making qualitative data machine read-
able by richly representing it in the Cube ontology along with metadata, and it enables
human-oriented textual responses to be annotated with terms from the risk terminology.
This paper describes the following contributions:
• The Cube Ontology, ARK projects vocabulary, and ARK risk terminology. They
enable formal and structured socio-technical analysis of systems and SKOS rep-
resentation of the background knowledge on risk management (including prove-
nance of term definitions).
• The ARK application. This provides a user interface for curation of struc-
tured Cube responses (including interlinking supporting evidence) and naviga-
tion around the multi-dimensional Cube system analysis or risk mitigation project
analysis.
Crotti Junior, A. et al. /
• Case study-based evaluation. We present preliminary results of a case study where
the ARK Platform has been used for Cube socio-technical analysis in a real-world
problem.
The remainder of this paper is organised as follows. A motivating use case is pre-
sented in Section 2. Section 3 describes the development of the ARK Platform. Section 4
presents a case study and preliminary results. Section 5 discusses related work. Section
6 concludes the paper.
2. Use Case
This section describes background on the Cube methodology and the high level use case
and requirements for risk management tools.
2.1. Background on the Cube methodology
The Cube methodology has been developed over twenty years by organisational psy-
chologists working on safety systems for the health, aviation and finance domains. It
supports progressively building a multi-dimensional analysis of complex systems [1]. It
provides a framework for initiating and analysing organizational change through inter-
linked, dimension-specific questionnaires. The general Cube questionnaire has 96 major
questions systematically organised into six project stages (Problem, Solution, Develop,
Implement, Plan and Prepare, and Verify and Embed), four aspects (Activity, Culture,
Functional System and Sensemaking), and four dimensions (Goal, Process/Sequence,
Social, and Information and Knowledge). The methodology guides an expert probing an
organisation using questions and then analysing it by systematic combination of sets of
related three-dimensional answers to identify risks, mitigations, and sources of value.
Applying the Cube methodology is a continuous process of improvement of an organiza-
tional system. Risks and mitigations are identified and this initiates a new improvement
project which is evaluated and verified. At each stage, the experts will fill out questions
and complete analyses.
2.2. Early Cube Tool Support
Prior to our research, the Cube methodology was manually deployed by a risk manage-
ment domain expert using a cumbersome spreadsheet as an aid. Limitations of this ap-
proach included insufficient navigation flexibility as analysis often requires non-linear
completion of questionnaires, a lack of embedded rules or relations between question an-
swers, inability to easily generate different interactive views of the data for analysis, no
ability to support for semi-automated analysis or validation of answers, no interlinking
to existing datasets, analytics or evidence, limited ability to reuse analysis results from
project to project, no sharing or team working support and no embedded reference glos-
sary of terms for safety analysts. A fair characterisation of this system would be to say it
was only suitable for highly trained domain experts and provided no automation or data
synthesis supports.
Crotti Junior, A. et al. /
2.3. Clinical Risk Management
Risk management is one of the most important procedures performed within clinical
environments, with strong impacts on the quality of service [4]. Due to the high volume
of available risk data from different sources (e.g. incident reports, electronic healthcare
records, best practice guidelines), it is often impractical in terms of time and effort for
humans to reliably extract the most relevant risk information from the data and use it for
effective decision-making and process improvement recommendations. Accordingly, it is
urgent that we develop innovative semi-automated but fair, accountable, and transparent
solutions to fuse, summarize, and categorise the risk data and evidence into patterns and
knowledge that will help safety experts to more effectively address safety improvement
in an evidence-driven way.
3. The ARK Platform
In this section, we present the Access Risk Knowledge (ARK) platform and ontolo-
gies that support socio-technical analysis of clinical risk management using the Cube
methodology. The ARK Platform relies on Semantic Web technologies and W3C stan-
dards to integrate interoperable qualitative clinical risk management data with quantita-
tive operational data and analytics. The platform is composed of the Cube ontology2 and
the Project vocabulary3 which is used to support data capture and analysis through the
Cube methodology, a safety management taxonomy4 described using W3C’s standard
Simple Knowledge Organisation System (SKOS)5 , and a prototype web-based applica-
tion, which allows domain experts to apply social-technical analysis through the Cube
methodology. Figure 1 presents an overview of the the ARK platform.
Figure 1. The ARK Platform overview
2 Available at https://w3id.org/ARK/Cube.
3 Available at https://w3id.org/ARK/Projects.
4 Available at https://w3id.org/ARK/Terminology.
5 https://www.w3.org/2004/02/skos/.
Crotti Junior, A. et al. /
3.1. The Cube Ontology and Project vocabulary
The Cube ontology and the Project vocabulary were developed to enable data capture and
analysis through the Cube methodology. The ontologies are based on two years of iter-
ative development between mixed teams of organisational psychologists and knowledge
engineers. The ontologies were implemented using the Web Ontology Language (OWL)
specification6 in protégé7 . Moreover, the ontologies were designed following Gruber’s
[5] principles and guidelines for ontology development. The purpose of such principles
is to assure knowledge sharing, reusability and interoperability. Thus, the ontologies con-
tain human-readable metadata as well as documentation, are free of inconsistencies, are
extensible and customizable allowing for different Cube questionnaires to be instanti-
ated while still following concisely the multidimensionality of the Cube methodology.
The Cube ontology and Project vocabulary include major concepts, structures and rela-
tionships together with a concise definition of the terminologies described in the Cube
methodology. The ontologies have also been validated using the Ontology Pitfall Scanner
[6], documented with machine and human readable metadata, which is used to automat-
ically generate documentation through WIDOCO [7], and are served following Linked
Data principles. Figure 2 presents the core components of the Cube ontology, which are
discussed in the following subsections.
Figure 2. Core components of the Cube ontology
The Cube ontology is used in the overall architecture of the ARK Platform in two
ways. First, it serves as a repository of the standard Cube questionnaire. Moreover, when
there is need for customization, depending on the domain or organization where the Cube
methodology is being applied, different Cube questionnaires can be instantiated. Second,
6 https://www.w3.org/TR/owl2-primer/
7 https://protege.stanford.edu/
Crotti Junior, A. et al. /
the Cube ontology is used to represent responses to the Cube questionnaire, enabling
an evidence base that supports governance within the organization. On the other hand,
the Project vocabulary allows organizations to structure risk analysis through the Cube
methodology by way of projects and sub-projects.
3.2. Safety Management Taxonomy
The safety management taxonomy was defined using SKOS, a W3C Recommendation
designed to support the use of knowledge organization systems. In the ARK Platform,
the developed taxonomy allows users to annotate risk management data described using
the Cube ontology with qualitative operational risk data. Concepts in the taxonomy have
been defined by domain experts and have at least a description and, when available, their
source. Since these domain experts do not have expertise in Semantic Web technologies,
the taxonomy was defined in a collaborative environment which is later transformed.
This transformation was performed using R2RML8 mappings created with the aid of an
editor [8]. Finally, the taxonomy was documented using WIDOCO and, like the Cube
ontology and Projects vocabulary, is served following Linked Data principles.
3.3. ARK application
The ARK application was built using Node.js and React, and offers three main interfaces
where users can interact with the Cube methodology. The first component allows orga-
nizations to define projects. The second component (Figure 3) guides users into using
the Cube methodology by allowing questionnaires to be answered. In order to provide
experts with a more comprehensive and granular analysis, this component allows users
to interlink answers with supporting evidence. Finally, this component also provides the
functionality to classify answers with risk-related concepts using the developed safety
management taxonomy.
Figure 3. Cube analysis interface: answering Cube questions.
The third component, called Cube summary, focuses on the summarization and vi-
sualisation of answers at the end of each stage. The Cube summary component provides
8 A a W3C Recommendation allowing one to define customized mappings to convert non-RDF resources to
RDF. https://www.w3.org/TR/r2rml/
Crotti Junior, A. et al. /
functionality that enables viewing answers similarly to a Rubik’s Cube, where the vi-
sualisation of any stage, aspect and dimension can be configured by the user. Figure 4
presents the Cube summary component.
Figure 4. Cube summarization: dynamic summary of the cube analysis
4. Case Study
The ARK platform has been developed with the aim of facilitating risk analysis of com-
plex systems through the Cube methodology. To evaluate the platform, a preliminary
case study has been conducted, in which, two participants were asked to use the plat-
form to address a clinical risk management problem and report their experiences through
separate interviews.
Participants were recruited and introduced to the case study by one of the research
team members. To assess the different levels of expertise, one of the participants was an
expert in risk management, while the other was a non-expert user.
4.1. Experiment Task: The Clinical Risk Management Case
The task participants were asked to perform was related to a clinical risk management
case which had previously been described in an MSc thesis [9]. The thesis analysed
the impact of collective leadership approaches on the integration of electronic health
record systems. Participants were asked to study the aforementioned thesis, and analyse
its risk in change using the ARK Platform. It is important to note that this is the first time
participants had access to this material. Participants worked on this task for 4 days.
Crotti Junior, A. et al. /
4.2. Discussion
After task completion, participants were interviewed in relation to their reflection over
its use, integrity/consistency of functions in the platform, and efficacy. Other features
evaluated include:
• Descriptions of concepts defined in the safety management taxonomy concepts
within the project analysis interface.
• The ability to navigate questions and answers in the ARK platform using different
dimensions and aspects.
• The ability to hide features/aspects.
• The ability to summarize user’s answers.
• The ability to provide evidences to qualitative answers.
Participants found training to be necessary when being introduced to the platform.
Nonetheless, participants evaluated its learning curve as fairly easy. Participants de-
scribed the platform as complex in relation to the amount of information needed in each
question, and the number of questions to be answered. Participants also described that the
large number of questions within the Cube questionnaire contributed to a comprehensive
understanding of the complex system presented in the experiment task. In this context,
even though the summary component was found useful, participants described having to
switch back and forth through different dimensions and aspects continuously. Further-
more, participants also described the use of platform as time-consuming (estimated at 3
to 4 hours to complete). In terms of integrity, the platform was verified as consistent, in
which, the relation between different functionalities is clear and easy to navigate. Other
features, such as the possibility of interlinking data sources as evidences, taxonomy def-
initions being available through the interface, the ability to hide features/aspects, have
also been considered helpful.
Generally, in terms of efficacy, participants found the platform useful and they con-
firmed that it gives a comprehensive overview of the analysed use case. The reaction over
Semantic Web technologies-driven features to support the Cube methodology has had a
positive outcome. Even though our preliminary results are promising, a comprehensive
evaluation using other instruments, such as standard usability tests and a larger number
of participants, is required to validate our findings.
5. Related Work
The use of ontologies for the analysis of organisational change and risk management
has been proposed in different domains. For instance, the authors in [10] used ontology-
based risk management framework to enhance the management of organisational risk.
Ontology-based feed-forward risk management is also proposed in [11] to create a multi-
level approach for feed-forward risk management. Other ontology applications are used
to support risk assessment in supply networks [12]. A more specific application of
ontology-based risk management is proposed in [13] which focuses on the development
of a risk knowledge management integrated into the Business Information Modelling
environment for construction risk knowledge management. The risk knowledge is rep-
resented using an ontology to annotate documents and produce a map which is used to
Crotti Junior, A. et al. /
infer paths. In contrast, our work is focused on supporting change risk analysis processes
using ontologies, Semantic Web and Linked Data technologies by leveraging the Cube
methodology.
There are also ontology development efforts towards the semantic modelling of sys-
tematic data collection and analysis tools such as questionnaires. The semantic mod-
elling of traditional questionnaires [14, 15] allows the semantic representation of ques-
tionnaires and their answers. Even if their work focuses on a traditional data collection
of linguistic importance and on the semantic enrichment and interlinking of manually
collected data, the structure used in these ontologies overlap with the structure of the
Cube questionnaire. These ontologies, however, are not expressive enough to represent
the Cube methodology..
6. Conclusion and Future Work
In this paper, we introduce the ARK platform and ontologies for socio-technical analy-
sis applied to clinical risk management. The proposed platform provides an extensible
interface, that is customizable in terms of its representation, and interoperable through
its ontology-driven back-end for representing social-technical analysis through the Cube
methodology. Our preliminary evaluation of the system shows that our proposed platform
is capable of representing core concepts, relationships, customisation, and visualisation
of the Cube methodology. A limitation of our evaluation is the number of participants
which will be targeted in future experiments.
Future work includes enhancing the platform with additional functionalities includ-
ing rich domain knowledge related to organisational risk, knowledge extraction from the
supporting documents, and improvements described by participants of our preliminary
study. Finally, future work will also be focused on evaluating the platform with a wide
community of users through standard tests.
Acknowledgements
This research was conducted with the financial support of Enterprise Ireland under Grant
Agreement No. CF 2018-2012 at the ADAPT SFI Research Centre. The ADAPT SFI
Centre for Digital Media Technology is funded by Science Foundation Ireland through
the SFI Research Centres Programme and is co-funded under the European Regional
Development Fund (ERDF) through Grant # 13/RC/2106. The research team would like
to thank participants of the case study for their help with the ARK platform evaluation.
References
[1] Siobhán Corrigan, Alison Kay, Katie O’Byrne, Dubhfeasa Slattery, Sharon Shee-
han, Nick McDonald, David Smyth, Ken Mealy, and Sam Cromie. A socio-
technical exploration for reducing & mitigating the risk of retained foreign objects.
International Journal of Environmental Research and Public Health, 15(4), 2018.
ISSN 1660-4601. doi: 10.3390/ijerph15040714. URL https://www.mdpi.
com/1660-4601/15/4/714.
Crotti Junior, A. et al. /
[2] Daniele Baranzini. Features of unstable approach in aviation: Big data 2.0 evidence
2018. 04 2018. doi: 10.13140/RG.2.2.34382.15686.
[3] Nick Mcdonald. The evaluation of change. Cogn. Technol. Work, 17(2):193–206,
May 2015. ISSN 1435-5558. doi: 10.1007/s10111-014-0296-9.
[4] Louise Dineley. Risk management systems in healthcare. PhD thesis, University of
Kent, 2016.
[5] Thomas R Gruber. Toward principles for the design of ontologies used for knowl-
edge sharing? International journal of human-computer studies, 43(5-6):907–928,
1995.
[6] Marı́a Poveda-Villalón, Asunción Gómez-Pérez, and Mari Carmen Suárez-
Figueroa. OOPS! (OntOlogy Pitfall Scanner!): An On-line Tool for Ontology Eval-
uation. International Journal on Semantic Web and Information Systems (IJSWIS),
10(2):7–34, 2014.
[7] Daniel Garijo. Widoco: a wizard for documenting ontologies. In In-
ternational Semantic Web Conference, pages 94–102. Springer, Cham, 2017.
doi: 10.1007/978-3-319-68204-4 9. URL http://dgarijo.com/papers/
widoco-iswc2017.pdf.
[8] Ademar Crotti Junior, Christophe Debruyne, and Declan O’Sullivan. Juma up-
lift: Using a block metaphor for representing uplift mappings. In 12th IEEE In-
ternational Conference on Semantic Computing, ICSC 2018, pages 211–218. doi:
10.1109/ICSC.2018.00037.
[9] Arwa Shuhaiber. Collective leadership for healthcare system change and clinical
collaboration, 2017.
[10] H. Ping Tserng, Samuel Y.L. Yin, R.J. Dzeng, B. Wou, M.D. Tsai, and W.Y. Chen.
A study of ontology-based risk management framework of construction projects
through project life cycle. Automation in Construction, 18(7):994 – 1008, 2009.
ISSN 0926-5805. doi: https://doi.org/10.1016/j.autcon.2009.05.005.
[11] Cédric Gaspoz, Ulysse Rosselet, Mathias Rossi, and Mélanie Thomet. Ontology
driven feedforward risk management. In Álvaro Rocha, Hojjat Adeli, Luı́s Paulo
Reis, and Sandra Costanzo, editors, New Knowledge in Information Systems and
Technologies, pages 253–261, Cham, 2019. Springer International Publishing.
ISBN 978-3-030-16181-1.
[12] Claire Palmer, Esmond N. Urwin, Ali Niknejad, Dobrila Petrovic, Keith Pop-
plewell, and Robert I. Young. An ontology supported risk assessment approach for
the intelligent configuration of supply networks. J. Intell. Manuf., 29(5):1005–1030,
June 2018. ISSN 0956-5515. doi: 10.1007/s10845-016-1252-8.
[13] L.Y. Ding, B.T. Zhong, S. Wu, and H.B. Luo. Construction risk knowledge man-
agement in BIM using ontology and semantic web technology. Safety Science, 87:
202 – 213, 2016. ISSN 0925-7535. doi: 10.1016/j.ssci.2016.04.008.
[14] Yalemisew Abgaz, Amelie Dorn, Barbara Piringer, Eveline Wandl-Vogt, and Andy
Way. Semantic modelling and publishing of traditional data collection question-
naires and answers. Information, 9(12), 2018. ISSN 2078-2489. doi: 10.3390/
info9120297.
[15] Yalemisew M. Abgaz, Amelie Dorn, Barbara Piringer, Eveline Wandl-Vogt, and
Andy Way. A semantic model for traditional data collection questionnaires en-
abling cultural analysis. In Proceedings of the Eleventh International Conference
on Language Resources and Evaluation (LREC 2018), 2018.